diff options
| author | David Gibson <david@gibson.dropbear.id.au> | 2026-02-04 21:41:34 +1000 |
|---|---|---|
| committer | Stefano Brivio <sbrivio@redhat.com> | 2026-02-25 00:17:27 +0100 |
| commit | e48ce41a1ec2f05846fb66d3847c2c2b6448ca71 (patch) | |
| tree | 608473d101cf513f81026575c39fadcbcd55d4e6 /test/lib | |
| parent | eb3babf7b4cf3793e0ed48030db958ace6546d7e (diff) | |
| download | passt-e48ce41a1ec2f05846fb66d3847c2c2b6448ca71.tar passt-e48ce41a1ec2f05846fb66d3847c2c2b6448ca71.tar.gz passt-e48ce41a1ec2f05846fb66d3847c2c2b6448ca71.tar.bz2 passt-e48ce41a1ec2f05846fb66d3847c2c2b6448ca71.tar.lz passt-e48ce41a1ec2f05846fb66d3847c2c2b6448ca71.tar.xz passt-e48ce41a1ec2f05846fb66d3847c2c2b6448ca71.tar.zst passt-e48ce41a1ec2f05846fb66d3847c2c2b6448ca71.zip | |
tcp: Remove non-working activity timeout mechanism
This mechanism was intended to remove connections which have had no
activity for two hours, even if they haven't closed or been reset
internally. It operated by setting the two hour timeout if there are
no sooner TCP timeouts to schedule.
However, when the timer fires, the way we detect the case of the activity
timeout doesn't work: it resets the timer for another two hours, then
checks if the old timeout was two hours. But the old timeout returned
by timerfd_settime() is not the original value of the timer, but the
remaining time. Since the timer has just fired it will essentially always
be 0.
For now, just remove the mechanism, disarming the timer entirely if there
isn't another upcoming event. We'll re-introduce some sort of activity
timeout by a different means later.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
Diffstat (limited to 'test/lib')
0 files changed, 0 insertions, 0 deletions
